Introduction to RMarkdown
RMarkdown is a powerful “literate programming” environment, meaning that one can easily combine text, code and output - including interactive htmlwidgets. There are three types of document that can be created:
- PDF: Useful when you want to print documents (and remains the default output format in academic publishing)
- MS Word: Useful if others you’re working with need to work with MS Word documents
- HTML: Useful for creating interactive content you can publish to the web
All HTML documents (including reports and presentations) created with RMarkdown can be published freely (and publicly) to RPubs.com.
All of the content below uses the same code as from the cheats_data-viz.R file but demonstrates how different types of content can be combined. The data behind these examples was extrapolated and anonymised from the following Figshare deposit: https://doi.org/10.6084/m9.figshare.4516772.
The following packages are used in this document, and all data is from the data/cheats_journeys.csv file.
library("tidyverse")
library("leaflet")
library("highcharter")
library("sf")
library("statesRcontiguous")
library("lubridate")
journeys <- read_csv("data/cheats_journeys.csv")Maps
The leaflet library allows us to create interactive maps, here are two common GIS visualsiations. To learn morw about GIS visualisations, please do visit the Oxshef Charts website.
Scatter geo plots
scatter geo plots are very useful for simply demonstrating where events occured. In the map below circle markers are used to show the start and end locations in red and green, respectively.
journeys %>%
leaflet() %>%
addTiles() %>%
addCircleMarkers(
lng = ~ start.longitude,
lat = ~ start.latitude,
color = "red",
radius = 1
) %>%
addCircleMarkers(
lng = ~ end.longitude,
lat = ~ end.latitude,
color = "green",
radius = 1
)Choropleth
The leaflet library allows us to easily visualise choropleth, provided we have appropriate shapefiles available. Here we prepare a shaepfile of the contiguous USA using the statesRcontiguous library:
contiguous_usa <- shp_all_us_states %>%
filter(contiguous.united.states == TRUE)Using the sf library it is fairly easy to calculate the number of journeys that started in each State:
state_send_locs <- journeys %>%
filter(start.country == "USA") %>%
st_as_sf(
coords = c("start.longitude", "start.latitude"),
crs = st_crs(contiguous_usa)
)
contiguous_send_counts <- contiguous_usa %>%
mutate(send.counts = st_covers(contiguous_usa, state_send_locs) %>%
lengths())Finally, we can create a palette and visualise this data:
palette_contiguous_us <-
colorBin("YlOrBr", domain = contiguous_send_counts$send.counts)
contiguous_send_counts %>%
leaflet() %>%
addPolygons(
fillColor = ~ palette_contiguous_us(send.counts),
fillOpacity = 1,
weight = 1,
color = "#000",
label = ~ paste0(state.name, " (", send.counts, " journeys)")
) %>%
addLegend(pal = palette_contiguous_us,
values = ~send.counts)Calendar heatmap
ggplot2 is an extraudinarily powerful and versatile tool for visualising data with R, implementing a consistent and complete “grammar of graphics”. As an example of how far you can go with ggplot2 this is a calendar heatmap showing the distribution of letters sent through the 1860s in the journeys dataset.
dated_journeys <- journeys %>%
select(date, number.of.letters) %>%
mutate(year = year(date),
yearmonthf = paste(month(date, label = TRUE), year(date)),
monthf = month(date, abbr = TRUE, label = TRUE),
week = week(date),
monthweek = ceiling(day(journeys$date) / 7),
weekdayf = wday(date,label = TRUE))
gg_dated_journeys <- dated_journeys %>%
filter(date >= dmy("01-01-1860") & date <= dmy("01-12-1869")) %>%
ggplot(aes(monthweek, weekdayf, fill = number.of.letters)) +
geom_tile(colour = "white") +
facet_grid(year~monthf) +
scale_fill_gradient(low="red", high="green") +
labs(x="Week of Month",
y="",
title = "Number of letters sent per day in the 1860s",
fill = "Number of letters")
gg_dated_journeys
Comparing ggplot2 and highcharter
The highcharter library allows us to create very professional looking interactive charts and plots. It’s important to note that the library is NOT free to use for commercial or governmental usage, though we can use it when communicating research outputs.
Both ggplot2 and highcharter use a very similar syntax:
ggplot(data, aes(x = x, y = y)): theaesfunction sets the aesthetics for the ggplot object, which columns indatashould be used for which visual properties of the chart.hchart(data, hcaes(x = x, y = y)): thehcaesfunction sets the aesthetics for the highcharter object, which columns indatashould be used for which visual properties of the chart
Country -> Country journeys tally
If you refer to the .Rmd file used to generate this report, you’ll notice the section header has {.tabset} appended. This allows us to create the tabbed content below from any child subheading of the current heading level.
ggplot2
country_to_country_counts <- journeys %>%
count(start.country, end.country) %>%
mutate(journey = paste(start.country, "->", end.country)) %>%
arrange(n) %>%
mutate(journey = as.factor(journey)) %>%
mutate(journey = fct_reorder(journey, n))
gg_country_to_country_counts <- country_to_country_counts %>%
ggplot(aes(x = journey, y = n)) + geom_col() +
coord_flip() +
xlab("") +
ylab("Number of journeys") +
ggtitle("Number of journeys split by start and end country")
gg_country_to_country_counts
highcharter
journeys %>%
count(start.country, end.country) %>%
mutate(journey = paste(start.country, "->", end.country)) %>%
arrange(desc(n)) %>%
hchart(type = "bar",
hcaes(x = journey, y = n)) %>%
hc_xAxis(title = list(text = "")) %>%
hc_yAxis(title = list(text = "Number of journeys")) %>%
hc_title(text = "Number of journeys split by start and end country")Grouped bar charts
When constructing grouped or stacked bar charts in ggplot2 or highcharter one must ensure to reshape data into long format:
- variable column: a measure of the data
- value column: a value for a specific measure of the data
This is achieved with gather from the tidyr library.
end_country_tallies <- journeys %>%
group_by(end.country) %>%
summarise(total.letters = sum(number.of.letters),
total.journeys = n()) %>%
mutate(
total.letters = total.letters / sum(total.letters),
total.journeys = total.journeys / sum(total.journeys)
) %>%
arrange(total.letters) %>%
mutate(end.country = fct_reorder(end.country, total.letters)) %>%
gather(measure, value,-end.country)
end_country_tallies## # A tibble: 12 x 3
## end.country measure value
## <fctr> <chr> <dbl>
## 1 POL total.letters 2.129971e-05
## 2 BEL total.letters 2.108671e-03
## 3 USA total.letters 2.268419e-02
## 4 GDR total.letters 1.583633e-01
## 5 DEU total.letters 1.875013e-01
## 6 GER total.letters 6.293212e-01
## 7 POL total.journeys 1.163467e-03
## 8 BEL total.journeys 5.235602e-03
## 9 USA total.journeys 9.482257e-02
## 10 GDR total.journeys 3.606748e-02
## 11 DEU total.journeys 3.030832e-01
## 12 GER total.journeys 5.596277e-01ggplot2
end_country_tallies %>%
ggplot(aes(x = end.country,
y = value,
fill = measure)) +
geom_col(position = "dodge") +
coord_flip() +
xlab("Final Destination Country") +
ylab("Percentage") +
ggtitle("Final destination of letters",
subtitle = "Ordered by percentage of journeys") +
scale_y_continuous(labels = scales::percent) +
scale_fill_manual(
values = c("#1b9e77", "#7570b3"),
name = "",
breaks = c("total.letters", "total.journeys"),
labels = c("Percentage of Journeys", "Percentage of letters")
)
highcharter
end_country_tallies %>%
hchart(type = "bar",
hcaes(x = end.country,
y = 100 * value,
group = measure)) %>%
hc_xAxis(title = list(text = "Final Destination Country"),
reversed = FALSE) %>%
hc_yAxis(title = list(text = "Percentage"),
labels = list(format = '{value}%')) %>%
hc_title(text = "Final destination of letters") %>%
hc_subtitle(text = "Ordered by percentage of journeys")